Seminar on Integration of Quantum Computing Accelerators with HPC Systems (IN2107)

Prof. Dr. Martin Schulz, Muhammad Arslan AnsariAmr Elsharkawy, Jonas Winklmann, Xiaorang Guo

Dates:

Pre-course meeting :

  • Date: 08/02/2023
  • Time: 10:00 AM
  • Location: 02.08.020

Access the matching system: 09.02. to 14.02.2023
Matching system results: 23.02.2023
Topics assignment: [24.02. to 10.03.2023] 

Planning meeting:

Between 25.01.2023 and 08.02.2023
First meeting: To be decided
ECTS: 5
Language: English
Type: Seminar, 2S
Moodle course:  
Registration: Matching System

Motivation

Quantum applications operating as production software is yet to be fully developed. For such applications it is inevitable that they will partially run on quantum computers and classical computers. For applications where the development of a quantum application to solve the computational problems are a potential solution, regular high performance computer applications often already exist. Therefore, the natural progression is to have the classical and the quantum algorithms work together, which implies both classical and quantum computers will interact, as opposed to working in isolation. At the moment, such interaction, most of the time, happens remotely over the cloud vis REST API calls. Yet, as quantum processing units (QPUs) become more widely accessible, lower-level system software infrastructure is essential to have tighter integration and co-processor programming model. The toolchain target is to map the high level hybrid quantum-classical source code all the way to the instruction set architecture (ISA) which gets executed on both classical CPUs and integrated QPUs. Moreover, as quantum computing develops, having a clear interface between software and hardware, i.e. the ISA, can provide benefits. A properly designed ISA can offer a compact and efficient way to utilize the unique capabilities of a computing device, which can enable branch and feedback support. Having a general and efficient toolchain faces several challenges, one of the main challenges is that different QPU hardware technologies (e.g. superconducting QPU, trapped ions QPU, neutral atoms QPU, etc) require completely different approaches and techniques to be be handled efficiently.

The seminar goal is to dive into the existing toolchains, ISAs and unique algorithms used to realize efficient QC-HPC integration process. The list of potential topics to be covered includes (but is not limited to):

  • QC-HPC compilation toolchain (front-end, middle layer, and back-end) (paper)
  • Atom sorting algorithms (paper)
  • Quantum instruction set architecture (paper
  • Quantum control processor - microarchitecture (paper)